Research On Train Positioning Based On Track Feature Matching And INS Integration

Train positioning is the foundation for the train control system to form effective control strategy and ensure safe and efficient railway operation.Advanced rail transit requires train positioning system to reduce the dependence on trackside equipment and have high accuracy,high reliability and high maintainability.Matching positioning is an effective method to improve the accuracy and reliability of train positioning.To improve the robustness and accuracy of the train positioning system,it is of great practical significance to find a high-precision,reliable and low-cost matching positioning signal,which does not need the auxiliary trackside equipment.To meet the requirement of accurate and reliable autonomous train positioning for advanced rail transit,this thesis proposes a multi-source fusion positioning method based on track irregularity matching and INS.Track irregularity is a ”harmful signal”that should be avoided for train operation safety and ride comfort.When it exceeds a certain threshold,it is necessary to adjust the track to the designed smoothness.However,these ubiquitous track irregularities can be regarded as the ”fingerprint” characteristics of the rails,which are available signal resources for train positioning.This thesis explores the positioning potential of the track irregularity.Inertial navigation system(INS)is used to sense and measure track irregularity to achieve matching positioning,and the matched position can be used to correct the INS.This track irregularity matching/INS integrated navigation scheme can achieve autonomous,all-weather,continuous and reliable positioning in the whole region without trackside equipment.The main research works and contributions of this thesis include the following:1.The feasibility of using track irregularity as a new signal source for accurate matching positioning is studied and demonstrated.Firstly,the equivalence between track irregularity and attitude measurement is analyzed,and it is demonstrated that track irregularity can be sensed by low-cost Micro-Electro-Mechanical-Systems(MEMS)INS installed on the train.Secondly,based on a large number of train experiment data,the track irregularity is verified from three aspects: time-frequency characteristics,spatial differentiation and time stability.The results show that the track irregularity has the characteristics of diverse specific frequency,feature stability and high differentiation,which has the potential to be applied for train matching positioning.Thirdly,typical interference factors of the track irregularity measuring and matching are studied,including train suspension system,train speed,running direction and sensor measurement error.The results show that the influence of these interference factors can be effectively solved or mitigated.These interference factors will not seriously degrade the performance of matching positioning.Finally,to solve the interferences of track adjustment or tamping operation and train motion state change,magnetic field feature matching is added as supplementary.The time-frequency characteristics,interference sources,challenging situation and the generations of these two signals are analyzed and compared.The results show that these two feature signals are complementary,which can be used together to improve the continuity and reliability of the matching positioning.2.The method of the track feature matching positioning and its accuracy level estimation is studied.Firstly,the generation and update of the track feature background map are studied as the basis of follow-up matching positioning.Secondly,based on the complementary characteristics of the track irregularity and the railway magnetic field,an integrated matching positioning scheme using these two features is designed.The results show that the positioning error is less than 1.40 m(99%)in the whole region,which significantly improves the continuity and reliability of the matching positioning.Finally,aiming at the traditional challenge that it is difficult to independently evaluate the accuracy level of the feature matching,this thesis proposes a new matching degree index,which can truly and effectively represent the actual matching accuracy level.Based on abundant train experiment data,the relationship between the matching degree index and the actual matching accuracy level is established,and the matching positioning error variance can be effectively estimated,which provides reliable variance of the observation noise for the follow-up feature matching/INS integrated positioning algorithm.3.Aiming at the accurate,reliable and continuous positioning requirements,a feature matching/INS integrated positioning scheme is designed and implemented.In this scheme,onboard IMU works as the core sensor for sensing and measuring the track irregularity as well as for inertial navigation.The matched position is used to correct the INS so as to form an integrated navigation based on a single sensor.This thesis compares the positioning performance in different configurations.(1)In the track irregularity matching/INS integrated positioning scheme,the positioning errors of a typical tactical grade INS and a MEMS INS are 1.25 m and 2.18 m(RMS),respectively.(2)Based on the above scheme,magnetic matching is added to enhance the positioning performance.The positioning errors of a typical tactical grade INS and a MEMS INS are effectively reduced to 0.15 m and 0.32 m(RMS),respectively.Based on the correction of the matched position,even MEMS INS can achieve accurate,reliable and autonomous train positioning with 1 m level of accuracy.(3)Since the train motion is strictly constrained on the railway track,Non-Holonomic Constraints(NHC)is added to further enhance the INS positioning performance.Different from previous works,it is found that the midpoint of carriage front and rear bogies(not the wheel or bogie)is the valid reference spot for NHC.Accordingly,NHC lever arm needs to be compensated to maximize the constraint effect of NHC.Compared with NHC without lever arm compensation,the lateral positioning error based on the proposed method reduces 46% in the absence of matched position observations.In summary,this thesis proposes a new positioning source ”track irregularity” and a multi-source fusion positioning method based on track feature matching and INS.This method can achieve long time and continuous train positioning without GNSS and trackside equipment.The research work of this thesis provides a new idea for train precision positioning,which can be worked as an effective supplement to enhance the accuracy and reliability of the current multisensory train positioning system.